Info
Whenever new nodes are being replaced, when old nodes are in an unhealthy state. You can imagine if the state is based on CPU/RAM, the nodes would be replaced when it’s too high and services are not responsive. This way, before the new Node is span up, the old one would be deleted and cut off, resulting in the inability of spinning new ones up.